Supplements
Mizutani, S., Pauwels, E., Stoven,V., Goto, S. and Yamanishi, Y.,
Relating drug-protein interaction network with drug side-effects
Figure S1: Molecular function annotations of proteins in enriched pathways
- Molecular function annotations of proteins in enriched pathways
Component-associated proteins in enriched KEGG pathway maps were characterized by 14 molecular function categories.
Left table counts:
(1) Column proteins extracted in the component
(2) proteins that contributed to the pathway enrichment
(3) proteins that were annotated with the molecular function categories
Right histogram shows the frequency of proteins within the molecular function categories.
Note that proteins annotated with more than one category were double-counted. The histogram shows that the majority of components contain proteins of different molecular function categories, such as "Ion channels", "G Protein-Coupled Receptors", and "Enzymes". This confirms that proteins of different molecular functions contributed to the pathway enrichment in most components.
Figure S2: ROC curve and Precision-Recall curve in the four methods.
- (a) ROC curve
- (b) Precision-Recall curve
The proposed approach, "SCCA Target protein" (blue) is compared against three other approaches, "OCCA Target protein" (red), "SCCA Chemical structure" (green), and "OCCA Chemical structure" (black), for prediction accuracy in (a) ROC curve and (b) Precision-Recall curve. For detailed description of each method, and the cross-validation scheme, see Section 4.3 "Performance evaluation".
Table S1: Extracted sets of proteins and side-effects and the associated drugs
- Extracted canonical components + high scoring drugs.
For targeted proteins
1st column: component number
2nd column: weight
3rd column: number of associated drugs
4th column: UniProt ID
5th column: protein annotation
For side-effects
1st column: component number
2nd column: weight
3rd column: number of associated drugs
4th column: side-effect term
For drugs
1st column: component number
2nd column: canonical component score
3rd column: number of features (proteins or side-effects) for each drug
4th column: PubChem compound ID
5th column: DrugBank ID
6th column: Drug Annotation
Table S2: Enriched (A) KEGG pathway maps and (B) KEGG BRITE terms in 80 canonical components
Table S3(A): Predicted side-effects based on the protein binding profiles for unchatecterized drugs in DrugBank
Table S3(B): Predicted side-effects based on the chemical substructure profiles for unchatecterized drugs in DrugBank
Softwares and Data
